Child Abuse & Neglect 76 (2018) 149–159

Contents lists available at ScienceDirect

Child Abuse & Neglect

journal homepage: www.elsevier.com/locate/chiabuneg

Research article

Temperamental sensitivity to early maltreatment and later family MARK

cohesion for externalizing behaviors in youth adopted from foster
care
⁎
Irene Tunga, , Amanda N. Noroñaa, Steve S. Leea, Audra K. Langleyb,
Jill M. Watermana
a
Department of Psychology, University of California, 1285 Franz Hall, Los Angeles, CA 90095, USA
b
Department of Psychiatry and Biobehavioral Sciences, University of California, 757 Westwood Plaza, Los Angeles, CA 90095, USA

AR TI CLE I NF O AB S T R A CT

Keywords: Children in foster care frequently have histories of physical/sexual abuse and neglect, increasing
Maltreatment their risk for externalizing behaviors (EB; e.g., aggression). According to the differential sus-
Temperament ceptibility theory, children with reactive temperaments (e.g., negative emotionality) may be
Externalizing behavior particularly vulnerable to early maltreatment, but may also benefit the most from environmental
Foster children
enrichment such as family cohesion. In a high-risk longitudinal sample of 82 children adopted
Adoption
from foster care in Los Angeles County from 1996 and 2001, we examined predictions of EB from
childhood to adolescence/young adulthood from temperament, preadoption maltreatment, and
adoptive family cohesion. Overall, results from generalized linear models and generalized esti-
mating equations (GEE) did not support differential susceptibility theory – specifically, youth
with early reactive temperament did not exhibit heightened sensitivity to maltreatment nor to
later adoptive family cohesion. Instead, reactive temperament was associated with higher EB at
initial adoptive placement and escalating EB across childhood, controlling for age, gender, race-
ethnicity, preadoption maltreatment, and adoptive family cohesion. Preadoption maltreatment
history was unrelated to baseline EB, although sexual abuse history predicted escalating child-
hood EB post-adoption, whereas exposure to family violence (e.g., domestic violence) inversely
predicted EB over time. By late adolescence/young adulthood 11–15 years post-adoption, rates of
arrest and substance use in this sample were relatively comparable to normative populations of
youth, although older age of adoption predicted more substance use in late adolescence/young
adulthood. Findings highlight early reactive temperament and preadoption maltreatment as
important risk factors to target for ameliorating patterns of EB growth in the first few years of
adoption.

1. Introduction

Children with a history of foster care are at heightened risk for numerous socio-emotional and behavioral difficulties, especially
youth externalizing behavior (EB) (e.g., aggression, delinquency) and related adult outcomes (e.g., substance problems) (Cutuli et al.,
2016; Leve et al., 2012; Vidal et al., 2017). Compared to the general population, male and female former foster youth are 4 and 13
times more likely, respectively, to be arrested for EB by the age of 21 years (Courtney et al., 2007), an alarming pattern given that EB

⁎
Corresponding author at: UCLA, 1285 Franz Hall, Box 951563, Los Angeles, CA 90095, USA.
E-mail address: itung@ucla.edu (I. Tung).

http://dx.doi.org/10.1016/j.chiabu.2017.10.018
Received 31 March 2017; Received in revised form 29 September 2017; Accepted 30 October 2017
0145-2134/ © 2017 Elsevier Ltd. All rights reserved.
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

itself predicts chronic criminality, substance use disorders, academic failure, and job/economic instability (Edwards, Gardner,
Hickman, & Kendler, 2016; Odgers et al., 2008). As a result, EB also contributes to substantial annual societal costs through law
enforcement and juvenile justice systems, prisons, rehabilitation programs, and hospitalizations (Welsh et al., 2008).
Although foster youth face significantly higher risk for EB across multiple stages of development, the specific developmental
pathways that lead to these long-term outcomes are not well understood. Efforts to identify targets for intervention are complicated
by the fact that foster youth often have histories characterized by significant stress (e.g., maltreatment) that can have enduring effects
on behavioral development. For example, adoption is conceptualized as a critical intervention for children in foster care, with meta-
analytic evidence that adopted children can display significant plasticity in their behavioral outcomes and “catch up” to the general
population (van IJzendoorn & Juffer, 2006). However, children also exhibit significant individual differences in responsiveness to
these environmental changes (Palacios & Brodzinsky, 2010). Variations in the protective effects of adoption may be influenced by
biologically-based traits, such as early temperament, that influence how children perceive and respond to their social environment, as
well as by the quality of parenting and family environment in the adoptive home. Understanding the developmental course of EB in
children adopted from foster care and identifying which pre- and post-adoption factors (including individual and environmental
factors) predict these outcomes across development is critical to designing targeted prevention and intervention programs that are
delivered during key developmental periods.

1.1. Maltreatment and EB development

Among the many risk factors foster children experience before adoption, one of the most pervasive is maltreatment, including
prenatal substance exposure and postnatal maltreatment such as physical and sexual abuse and/or neglect (Oswald, Heil, & Goldbeck,
2010). There are relatively few longitudinal studies of foster care youth, but postnatal maltreatment is a key predictor of sustained EB
over time (Simmel, 2007). Because maltreatment diverges sharply from the average expected environment, it is conceptualized as one
of the most toxic and severe environmental conditions for development (Rogosch, Oshri, & Cicchetti, 2010). Childhood maltreatment
initiates cascades of atypical development of neurobiological and physiological processes (e.g., HPA-axis functioning, amygdala
functional connectivity), emotion regulation, and the formation of attachment and healthy relationships (Cicchetti & Banny, 2014;
Rogosch et al., 2010). These effects can further differ based on subtypes of maltreatment (Kuhlman, Geiss, Vargas, & Lopez-Duran,
2015), ranging from physical and sexual abuse, to neglect and exposure to family violence (e.g., witnessing domestic violence
between caregivers or sibling abuse). Together, childhood maltreatment and EB exert substantial individual, family, and societal
consequences, and they contribute to the intergenerational continuity of psychopathology.

1.2. Temperamental sensitivity to the environment

Despite evidence that maltreatment disrupts development across multiple levels of functioning, not all children with maltreat-
ment histories exhibit EB, highlighting the need to identify individual and environmental factors that promote resilience (Haskett,
Nears, Ward, & McPherson, 2006). Although rarely examined in the foster care population per se, developmental psychopathology
studies more broadly have identified several biologically-based “vulnerability factors,” including temperament traits, that may dif-
ferentiate children most sensitive versus “resistant” to adversity. Following a diathesis-stress conceptualization of psychopathology,
children exposed to maltreatment or early harsh parenting who also had “difficult” or reactive temperaments (e.g., high negative
emotionality, high sensitivity, low frustration tolerance, low inhibition) were particularly at risk for developing EB and other psy-
chopathology (Kiff, Lengua, & Zalewski, 2011). That is, temperament acutely increased children's vulnerability to negative en-
vironments through Temperament x Environment interactions (Belsky, Hsieh, & Crnic, 1998).
Beyond sensitivity to early stress, emerging studies measuring a full range of caregiving environments have found that compared
to children with easy temperaments, children with reactive early temperaments may also benefit more from positive parenting
practices (Kochanska & Kim, 2013). That is, according to differential susceptibility theory, children with reactive temperaments may
be more sensitive to the environment, for better and for worse (Ellis, Boyce, Belsky, Bakermans-Kranenburg, & van IJzendoorn, 2011).
A growing literature consisting of cross-sectional, longitudinal, and emerging experimental studies support the plausibility that
reactive temperament may confer general heightened sensitivity to the social environment (Belsky & Pluess, 2009). For example, a
randomized controlled trial of infant-mother dyads found that highly irritable/reactive infants, who are traditionally considered at
risk for later EB, benefited more from a brief intervention designed to increase secure attachments compared to less irritable infants
(Cassidy, Woodhouse, Sherman, Stupica, & Lejuez, 2011). Although intervention studies examining patterns of temperamental sen-
sitivity to the environment are only just emerging, these preliminary studies suggest that compared to children with easy tem-
peraments, children with temperamental risk for EB may also show heightened environmental sensitivity to socially-based inter-
ventions.

1.3. Differential susceptibility and foster-adoptive youth

Although almost no studies of foster youth have investigated patterns of temperamental sensitivity to the environment in the
context of differential susceptibility, the implications of this hypothesis for children in foster care are significant: there are over
400,000 children in foster care in the US alone (U.S. Department of Health and Human Services, 2011). Children with complex
histories including maltreatment, who may present with more severe EB at initial placement, are particularly stigmatized and less
likely to be adopted into nurturing permanent homes (Leathers, Spielfogel, Gleeson, & Rolock, 2012). However, differential

150
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

susceptibility suggests that these same children more biologically vulnerable to maltreatment may also benefit the most from pla-
cement in a nurturing adoptive home, and perhaps even fare better than youth who appeared resilient to early adversity.
Importantly, in attempting to extend the implications of differential susceptibility to high-risk samples with complex histories
(e.g., foster-adoptive youth), several key developmental considerations must be considered. First, because early development re-
presents a sensitive period for brain and behavior plasticity (Knudsen, 2004), most differential susceptibility studies have focused on
children in infancy and toddlerhood (Hentges, Davies, & Cicchetti, 2015; Leerkes, Nayena Blankson, & O’Brien, 2009). Thus, it is
unclear if temperamental sensitivity to the environment extends beyond early childhood to influence youth adopted in later child-
hood or even adolescence. For example, infant negative emotionality moderated the association between child-care quality and later
adolescent EB, but the pattern of interaction was consistent with diathesis stress rather than differential susceptibility (Belsky & Pluess,
2012), suggesting that the effects of positive environments on temperamentally sensitive individuals may fade across development. In
contrast, temperamental sensitivity to early maltreatment may continue to negatively influence EB in adolescence (Rioux et al.,
2016).
Indeed, because most differential susceptibility studies are cross-sectional, it is unclear how early Temperament x Environment
interactions may influence later sensitivity to environmental changes. This is important, because many children at highest risk for EB
have already been exposed to maltreatment by the time interventions are implemented, and thus they also exhibit the most severe
initial EB. Severe initial EB is linked to treatment resistance (Masi et al., 2011), which suggests that children most sensitive to early
maltreatment (and thus with most severe initial EB) may not benefit the most from later intervention. When development is ade-
quately considered, it is unclear whether differential susceptibility applies to children already exhibiting severe EB due to their
vulnerability to early adversity. Furthermore, due to the naturalistic design of most EB studies, the source of maltreatment (i.e.,
caregivers) often continues to be present in the child's environment, making it difficult to distinguish between long-term effects from
early adversity versus effects from concurrent adversity. Thus, studies employing experimental or quasi-experimental designs that
explicitly measure changes in environmental experiences are poised to clarify developmentally-sensitive aspects of differential
susceptibility (van IJzendoorn & Bakermans-Kranenburg, 2012).

1.4. Study aims

In a high-risk sample of children removed from their biological parents and adopted from foster care, the present study had two
primary goals: First, we investigated the impact of early maltreatment (i.e., physical abuse, sexual abuse, neglect, exposure to family
violence) on EB development for children removed from maltreating environments and placed into adoptive homes. Given that this
model has been rarely examined longitudinally and across multiple stages of development for foster-adoptive youth, we tested how
subtypes of maltreatment predict children's levels of EB at initial placement into adoptive homes, patterns of EB change across the
first five years post-adoption, and long-term EB outcomes in adolescence/young adulthood.
The second primary goal was to explore the role of reactive temperament (e.g., negative emotionality) on these processes in the
context of emerging environmental sensitivity theories. In doing so, we explored the following research questions: (1) Does reactive
temperament represent a vulnerability factor for pre-placement risk (history of abuse or neglect) on EB levels at initial adoptive
placement? (2) Do these same temperament traits pose as sensitivity factors for post-placement adoptive family support to predict
decreases in later EB across the first five years of placement? And (3) beyond childhood outcomes, how might these Temperament x
Environment patterns change across time with respect to long-term EB outcomes such as arrest history and substance use? By
employing a study design that manipulated the environment through adoption as an intervention, this study aimed to provide
preliminary evidence to evaluate whether, as proposed by the differential susceptibility theory, youth with reactive temperaments are
at once more vulnerable to early maltreatment as well as benefit the most from measurable differences in adoptive family enrich-
ment.

2. Methods

2.1. Participants

Between 1996 and 2001, families of 82 children were recruited from the UCLA TIES for Adoption program (now TIES for
Families) to participate in a longitudinal study. TIES for Adoption aimed to facilitate successful adoption of high-risk children
transitioning from foster care to adoption. Based on requirements from the Adoptions Division of the Los Angeles County Department
of Child and Family Services (DCFS), prospective adoptive parents attended a series of educational seminars prior to being matched
with a child. Approximately 85% of these seminars included announcements about the adoption program, which offered three
additional educational meetings. Prospective parents who attended the three education meetings learned about available services
including pre-placement consultation, counseling services, and medical, educational, and psychiatric consultation. Families who
subsequently had children placed with them and requested services from TIES for Adoption were asked if they would like to par-
ticipate in our longitudinal research study. Children were eligible for the study if they were under 9 years of age at baseline and were
placed in the adoptive placement within the past 2 months.
Eligible participants completed seven separate assessments (baseline, 1-, 2-, 3-, 4-, and 5- years post-placement, and in a long-term
follow-up 11–15 years post-placement). Table 1 shows the descriptive statistics for all demographic and study variables. At baseline,
children ranged from 4 months to 8 years of age (average = 4 years), were about equal in the distribution of boys and girls, and
ethnically diverse: most children were Latino/a, Black, or Biracial, whereas 69% of adoptive parents were Caucasian. About 51% of

151
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

Table 1
Descriptive statistics of demographic and study variables (n = 82).

Variable M (SD) or% N

Age of adoption in years 3.92 (2.20) 82

Gender (% boys) 53.7% 82
Race-ethnicity 18.3% 82
Latino/a 32.9% 82
Black 26.8% 82
Caucasian 18.3% 82
Mixed or Other 22.0% 82
Reactive temperament 3.44 (.46) 74
Maltreatment history (any) 51.3% 81
Physical abuse 19.2% 81
Sexual abuse 3.8% 81
Exposure to family violence 7.7% 81
Neglect 30.8% 81
T2 Family cohesion 7.95 (1.27) 38
T1 Externalizing 58.63 (10.40) 64
T2 Externalizing 56.48 (12.80) 54
T3 Externalizing 55.65 (11.24) 52
T4 Externalizing 55.33 (11.43) 40
T5 Externalizing 59.44 (11.03) 36
T6 Externalizing 56.15 (11.58) 39
Arrest History 24.1% 54
Substance use 2.19 (1.52) 54

Note. T1, T2, etc. = timepoint 1, timepoint 2, etc.

youth had documented maltreatment history (i.e., physical or sexual abuse, neglect, or exposure to domestic violence or sibling
abuse). Participants without a documented maltreatment history were typically relinquished by their birth parents or removed from
their homes due to prenatal substance exposure of the child or the child's sibling in the home. Given the high-risk nature of this
sample, there was notable attrition that varied across time points, with families participating in some time points but not the others
(e.g., skipping one time point but returning to participate at later time points). Specifically, of the 82 total families participating at
baseline, the percentage of participants at each time point were as follows: T2 (1 year: 84%), T3(2 years: 76%), T4 (3 years: 52%), T5
(4 years: 52%), T6 (5 years: 51%), T7 (long-term follow-up: 66%). Across time points, 10% dropped out of the study after baseline; of
those who did not drop out, 81% (n = 60) of families participated in at least three follow-up time points. Although there were no
statistically significant differences (p's > 0.05) between participants and non-participants in the follow-up study on demographic
characteristics or study variables (i.e., maltreatment, temperament, family cohesion, EB), we implemented procedures to combat
missing data (described below).

2.2. Procedures

After determining eligibility, we obtained parental consent and mailed standardized rating scales to parents about their child's
temperament and behavioral and social-emotional functioning. When the child lived with two adoptive parents, the primary adoptive
parent (i.e., parent spending the most time with the child) completed the rating scales. In addition, the Department of Children and
Family Services (DCFS) granted permission to review the child's adoption records, which yielded additional data about the child's
preadoption history (e.g., abuse/neglect, medical and placement history). Maltreatment history data were collected and entered from
records by trained psychology graduate students under the supervision of the principal investigator. When records were unclear, the
principal investigator reviewed the charts to make final decisions about whether the information should be included. After the
baseline assessment, families returned each year for the next five years and completed highly parallel batteries including assessment
of the child's behavior and family functioning. Finally, a long-term follow-up was conducted after 11–15 years to assess the child's
outcomes in late adolescence. After obtaining parent and adolescent consent, parents and adolescents completed separate online
rating scales assessing emotional and behavioral functioning and related outcomes. All research procedures were approved by the
UCLA Institutional Review Board.

2.3. Measures

2.3.1. Temperament
At baseline, the primary adoptive parent completed the Cameron-Rice Temperament scales (Cameron & Rice, 1989), adapted from
previous youth temperament rating scales (Carey & McDevitt, 1978; Fullard, McDevitt, & Carey, 1984; Mcdevitt & Carey, 1978).
Developmentally parallel versions of the scales were administered based on the child's age at baseline: Infant Temperament Ques-
tionnaire for infants less than 1 years old (n = 4), Toddler Temperament Scale for toddlers 1–3 years old (n = 14), and the Pre-
school/Child Temperament Questionnaire for children older than 3 (n = 55). Parents rated items about their child's behavior on a
Likert scale from 1 (almost never) to 6 (almost always). The items formed seven subscales: sensitivity, activity level, reactivity/

152
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

intensity, frustration tolerance, adaptability, regularity, and soothability (infants only) or distractibility (toddlers and children only).
Three dimensions were unrelated with other temperament dimensions (i.e., frustration tolerance) or were age-specific and only
administered to a subset of participants (i.e., soothability and distractability). Excluding these three dimensions, a principal com-
ponent analysis (PCA) identified six subscales (low adaptability, negative mood, high sensitivity, high intensity, high activity, low
approach) that loaded onto a single dimension (loadings > 0.35) that was conceptually consistent with difficult or reactive tem-
perament. This composite measure consisted of the mean of these six subscales, with higher scores representing more reactive
temperament. The measure demonstrated predictive validity in this sample, significantly correlating with EB in expected directions
(r = 0.43, p < 0.01).

2.3.2. Preadoption maltreatment

The child's history of maltreatment (including physical abuse, sexual abuse, exposure to family violence, and neglect) was
gathered from the child's adoption records, including court and medical records and reports by social workers. Excluding neglect
based on prenatal substance exposure, 51% of the sample had at least one documented type of maltreatment, either in their birth
homes or in foster care. Four maltreatment variables were coded from court records as either present (coded as 1) or absent (coded as
0) for each child: (1) physical abuse, (2) sexual abuse, (3) exposure to family violence (i.e., directly witnessing sibling abuse or
domestic violence), and (4) neglect of basic needs (adequate supervision, food, clothing, shelter).

2.3.3. Post-placement family support

At the one-year follow-up time point in childhood, family support was assessed using an adapted version of the Family
Environment Scale (FES) (Moos & Moos, 1994). The adapted FES is a rating scale that includes 36 of the original 90 items and assesses
the interpersonal relationships and overall social environment of the family. The primary adoptive parent rated each item as True or
False, and items sum up (after reverse-scoring appropriate items) to form several subscales. To represent post-placement family
support, we used the Cohesion subscale (9 items, Mcdonald's ω = 0.63; Greatest Lower Bound (GLB) = 0.84), which included items
such as: “Family members really help and support one another” and “There is plenty of time and attention for everyone in our
family.” Previous studies reported predictive validity, such that family cohesion attenuated risk for delinquency and drug and alcohol
use among adolescents witnessing community violence (Barr et al., 2012; Kliewer et al., 2006).

2.3.4. Childhood EB
At baseline and at each of the five follow-up time points in childhood, the primary adoptive parent completed the Child Behavior
Checklist (CBCL), a standardized rating scale based on a normative sample of girls and boys that yields several broadband scales of
child symptomatology, including items related to EB such as aggressive behaviors (e.g., “gets in many fights,” “destroys things
belonging to his/her family or others”) and delinquency (e.g., “lying or cheating,” “runs away from home”) (Achenbach, 1991;
Achenbach & Rescorla, 2001). Parents rated each behavior based on the preceding 6 months as Not True (0), Somewhat or Sometimes
True (1), or Very True or Often True (2). We used T-scores from the Externalizing Behavior broadband scale at baseline and each
follow-up time point to model initial and prospective change in EB across time. The EB measure demonstrated excellent reliability
across the six childhood time points (McDonald's ω ranged from 0.90-0.93, GLB ranged from 0.89–0.94).

2.3.5. Long-term adolescent EB outcomes

At the long-term follow-up, parents and children (now late adolescents/young adults) separately answered questions about the
adolescent's behavior through parallel online questionnaires, adapted from items from the National Longitudinal Study of Adolescent
Health (Harris et al., 2008). Two long-term EB outcomes were assessed: arrest history and substance use. Arrest history was coded
dichotomously with 1 = ever been arrested or sent to juvenile hall (as reported by parent or child) and 0 = never arrested or sent to
juvenile hall according to both parent and child. Substance use ranged from 0 to 4 that represented the lifetime history of the number
of different substances the child used including cigarettes/tobacco, alcohol, marijuana, and “other illicit drugs”. A substance was
considered positively endorsed if self-reported by the adolescent or by the parent.

2.4. Data analytic plan

Three sets of analyses were conducted to predict (1) baseline EB, (2) time-varying childhood EB across the five follow-up time
points, and (3) long-term adolescent EB outcomes, respectively. All analyses were conducted in Stata 13.

2.4.1. EB at initial placement

First, to examine whether difficult temperament moderated the association between previous types of maltreatment and baseline
EB, we conducted a linear regression to predict baseline EB simultaneously from each type of maltreatment (physical abuse, sexual
abuse, exposure to family violence, and neglect), difficult temperament, and each maltreatment variable's interaction with tem-
perament. Child's age at adoptive placement, gender, and race-ethnicity were controlled as covariates to improve specificity of
effects. To increase power and address missing data issues due to attrition, we employed multiple imputation using 50 iterations of
multivariate imputation by chained equations (MICE) (Schafer & Graham, 2002). MICE employs an iterative multivariable regression
technique that treats each variable with missing data as an outcome variable and then imputes the missing values based on the
remaining variables in the model (in addition to relevant auxiliary variables). MICE has several advantages to other missing data
methods, including yielding more accurate standard errors by incorporating a random element to account for uncertainty in the

153
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

imputations, as well as appropriately accommodating non-normal data (e.g., count distributions) (Sterne et al., 2009).

2.4.2. Change in childhood EB

Next, to examine temperament and post-placement family cohesion as predictors of change in childhood EB after adoptive pla-
cement, we employed Generalized Estimating Equations (GEE) to model change in EB across the five childhood follow-up time points.
GEE is an extension of the general linear model that uses robust variance estimation in a repeated measures design (Hanley, Negassa,
Edwardes & Forrester, 2003). A significant advantage of GEE to general linear models is that it minimizes Type 1 error and increases
statistical power by adjusting for correlated observations across time points. Thus, the model tested the independent effects of family
cohesion (measured one year after adoption) and its interaction with early temperament on time-varying change in EB across the first
five years post-placement. We specified an exchangeable working correlation matrix; all tests were based on the z-statistic and ß
parameters are in logits. GEE addresses missing data using the “all available pairs” method, such that all non-missing pairs of data are
used in estimating the working correlation parameters. Thus, only the observation for that subject is missing rather than all variables
for that subject. We further employed Little's Test of Missing Completely at Random (Little, 1988) to evaluate whether individual
missing observations were missing completely at random (MCAR) and thus appropriate for listwise deletion. Little's test showed that
data were MCAR conditional on differences in race-ethnicity (i.e., covariate-dependent missingness; Li, 2013): χ2 (289) = 80.33,
p = 0.87. Thus, we controlled for race-ethnicity in the GEE as a covariate. To further enhance specificity of the family cohesion
effects, we also controlled for maltreatment history, age of adoption, and gender as covariates.

2.4.3. Long-term EB outcomes

Finally, to examine whether Temperament x Family Cohesion effects extended long-term into adolescence, we conducted two
complementary generalized linear models (GLMs) to separately predict arrest/juvenile-hall history and substance use. Specifically,
we employed logistic regression to predict adolescent arrest history and fit a generalized linear model to predict the number of
substances used, specifying poisson distributions to account for the skewed distribution of these count data. Each model consisted of
reactive temperament, adoptive family cohesion in childhood, and their interaction. Due to the smaller sample size at the long-term
follow-up time point, maltreatment types were combined into a cumulative “any history of maltreatment” variable and entered as a
single covariate to improve model fit. Adolescent's age, gender, and race-ethnicity were included as additional covariates when
significantly correlated with the outcome (i.e., arrest history or substance use). To reduce parameter biases associated with missing
data due to attrition, we employed multiple imputation using 50 iterations of multivariate imputation by chained equations (MICE)
for these models to increase power, accommodate the non-normal distributions, and attain more accurate standard errors
(Schafer & Graham, 2002).

3. Results

3.1. Preliminary analyses

Preliminary bivariate correlations showed that baseline EB was significantly (p < 0.05) and positively correlated with EB across
the first four years post-adoption (r = 0.65–0.75) as well as with arrest history (r = 0.37) and substance use (r = 0.32) 11–15 years
later. As expected, early reactive temperament was significantly and positively correlated with EB at baseline (r = 0.65), at the first
three years post-placement (r = 0.46–0.52), and with EB outcomes at the long-term follow-up (arrest history: r = 0.42; substance
use: r = 0.39), supporting the need to examine how reactive temperament may influence sensitivity to early maltreatment and later
family factors. Maltreatment subtypes showed small and inconsistent correlations with childhood EB that varied across time; whereas
physical abuse, sexual abuse, and neglect showed positive correlations with EB, exposure to family violence was negatively correlated
with EB at baseline (r = −0.34). Although adoptive family cohesion was not bivariately correlated with EB in childhood, it was
inversely correlated with arrest history in late adolescence (r = −0.74).

3.2. EB at initial placement

Results from the linear regression model predicting baseline EB are presented in Table 2. Controlling for child's age of adoption,
gender, race-ethnicity, and maltreatment history subtypes (physical abuse, sexual abuse, exposure to family violence, and neglect),
reactive temperament was significantly and positively associated with baseline EB. However, maltreatment subtypes and their in-
teractions with reactive temperament were not associated with baseline EB.

3.3. Change in childhood EB

Next, GEE modeled changes in EB across the childhood follow-up time points (Table 2). The associations between maltreatment
history and post-adoption EB differed based on subtype: sexual abuse was significantly associated with increasing EB across the first
five years post-adoption, whereas exposure to violence predicted decreasing patterns of EB across time. In contrast, physical abuse
and neglect were unrelated to EB change post-adoption after accounting for other covariates. Controlling for age at adoptive pla-
cement, gender, race-ethnicity, and the four maltreatment subtypes, reactive temperament was associated with increasing EB across
time, and it marginally moderated the association between family cohesion and changes in EB.

154
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

Table 2
Linear regression predicting baseline EB and generalized estimating equations predicting changes in EB across first five years of adoptive placement (n = 82).

Main Effects Adding Interaction

β (SE) p β (SE) p

Baseline EB (Linear Regression)

Age at Adoptive Placement 1.31 (0.77) 0.10 1.03 (.67) 0.13
Reactive Temperament 14.42 (2.93) < 0.01 15.12 (3.51) < 0.01
Maltreatment History
Physical Abuse −0.24 (2.74) 0.93 32.33 (23.90) 0.18
Sexual Abuse 4.54 (4.34) 0.30 28.42 (69.83) 0.69
Exposure to Family Violence −8.88 (5.31) 0.10 −45.84 (36.98) 0.23
Neglect .52 (2.43) 0.83 21.19 (23.70) 0.38
Temperament x Physical Abuse −9.23 (6.95) 0.19
Temperament x Sexual Abuse −5.17 (17.69) 0.69
Temperament x Family Violence 9.65 (9.57) 0.33
Temperament x Neglect −5.56 (6.69) 0.41
F-test 8.67 < 0.01 8.02 < 0.01

Changes in EB Post-Adoption (GEE)

Time < 0.01 (0.01) 0.77 < 0.01 (0.01) 0.79
Age at Adoptive Placement 0.01 (0.01) 0.33 0.01 (0.02) 0.42
Maltreatment History
Physical Abuse −0.04 (0.06) 0.44 −0.04 (0.06) 0.54
Sexual Abuse 0.20 (0.07) < 0.01 0.25 (0.07) < 0.01
Exposure to Family Violence −0.30 (0.07) < 0.01 −0.26 (0.08) < 0.01
Neglect −0.03 (0.05) 0.52 −0.01 (0.06) 0.81
Reactive Temperament 0.17 (0.07) 0.01 0.62 (0.25) 0.01
Family Cohesion 0.02 (0.03) 0.37 0.23 (0.11) 0.05
Temperament x Family −0.06 (0.03) 0.07
Wald chi2 39.00 < 0.01 46.43 < 0.01

Note. Significant effects boded for emphasis. Both models additionally included gender and race-ethnicity as covariates.

3.4. Long-term EB outcomes

When predicting arrest/juvenile hall history in the long-term follow-up, higher family cohesion was marginally associated with
lower risk for arrest history (Table 3). Maltreatment history was unrelated to arrest/juvenile hall contact in late adolescence, and
neither reactive temperament nor its interaction with family cohesion were associated with arrest/juvenile hall. When predicting the
number of substances used at the long-term follow-up, youth adopted at an older age used significantly more substances at the follow-
up. However, substance use was not associated with maltreatment history, reactive temperament, adoptive family cohesion, or the
interaction between temperament and family cohesion.

Table 3
Logistic and linear regressions predicting long-term EB outcomes from temperament, maltreatment history, and adoptive family cohesion (n = 82).

Main Effects Adding Interaction

β (SE) p β (SE) p

Arrest History
Maltreatment Historya −1.62 (1.87) 0.39 −1.81 (2.18) 0.41
Reactive Temperament 2.90 (2.52) 0.25 10.30 (13.39) 0.44
Family Cohesion −2.15 (1.17) 0.07 1.53 (5.30) 0.77
Temperament x Family Cohesion −1.11 (1.75) 0.53
F-test 1.85 0.14 1.76 0.14

Substance Use
Age at Adoptive Placement 0.11 (0.04) 0.02 0.11 (0.04) 0.02
Maltreatment Historya 0.23 (0.17) 0.17 0.22 (0.18) 0.22
Reactive Temperament 0.11 (0.18) 0.54 −0.19 (.86) 0.82
Family Cohesion −0.09 (0.06) 0.11 −0.24 (0.44) 0.59
Temperament x Family Cohesion 0.04 (0.12) 0.73
F-test 5.15 < 0.01 4.03 < 0.01

a
Maltreatment variable included any exposure to physical abuse, sexual abuse, neglect, and/or witnessing family violence. Significant main effects bolded for
emphasis.

155
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

Fig. 1. Conceptual diagram summarizing the associations between the independent variables and EB across time. Only statistically significant (p < 0.05; solid lines)
and marginally significant associations (p < 0.10; dotted lines) are shown. Note: a Maltreatment effects on childhood EB differed based on types of maltreatment (i.e.,
physical abuse, sexual abuse, neglect, exposure to violence); see Table 1 for the specific results.

4. Discussion

In a high-risk longitudinal sample of children adopted from foster care, we examined prospective predictions of EB across de-
velopment from preadoption maltreatment and early reactive temperament; we also explored whether temperament moderated
predictions of EB from early maltreatment and later adoptive family support. See Fig. 1 for a conceptual diagram summarizing the
primary findings across time points. First, controlling for age, gender, race-ethnicity, and reactive temperament, maltreatment history
(i.e., physical abuse, sexual abuse, exposure to family violence, neglect) was unrelated to baseline EB, but was associated with changes
in EB post-adoption. Specifically, sexual abuse predicted escalating EB in the first five years post-adoption, whereas family violence
exposure (i.e., witnessing sibling abuse or domestic violence) was associated with decreasing EB post-adoption. Second, reactive
temperament emerged as a consistent risk factor for EB in childhood, predicting higher EB at initial adoptive placement, as well as
increasing EB across the first five years of adoption. Children with reactive temperament were not more vulnerable to elevated EB
secondary to maltreatment, although they were marginally more sensitive to the protective effects of family cohesion during the first
five years post-adoption. Interestingly, by the adolescent/young adult follow-up (11–15 years after baseline), neither maltreatment
history nor temperament were associated with EB (i.e., arrest/juvenile hall, substance use), whereas adoptive family cohesion
continued to exhibit a marginally significant and protective effect on arrest history. Finally, although age of adoptive placement was
unrelated to childhood EB outcomes, it was inversely associated with later substance use, such that children adopted at a younger age
used significantly fewer types of substances by adolescence/young adulthood.

4.1. Maltreatment and EB development

To review, the first goal of this study was to examine the associations between preadoption maltreatment and EB across devel-
opment. Maltreatment was not associated with baseline EB, although it did influence patterns of change in EB during the first five
years post-adoption. That maltreatment was unrelated to baseline EB in this sample was somewhat surprising given prevailing
evidence (Jaffee, Caspi, Moffitt, & Taylor, 2004; Kerig & Becker, 2015). One potential reason for this divergence in findings is that this
study focused specifically on postnatal maltreatment (i.e., physical and sexual abuse, neglect). However, children removed from their
parents often experience multiple risk factors for EB beyond maltreatment while living in the biological home and/or afterwards in
foster care. Indeed, EB consists of numerous causal influences, including factors commonly experienced by youth removed from their
birth families, such as exposure to neighborhood violence, home instability (e.g., frequent moves), inconsistent parenting, and
parental psychopathology (Greeson et al., 2011; Healey & Fisher, 2011; Leve et al., 2012; Simmel, 2007). Because this study em-
ployed a clinical sample without a normative comparison group, the “non-maltreated” youth may have experienced these other
significant stressors in addition to prenatal forms of maltreatment such as prenatal exposure to teratogens. Relatedly, children
adopted from foster care can vary widely in the quality of foster care experienced, with some children experiencing more mal-
treatment during their foster placements (Euser, Alink, Tharner, van IJzendoorn, & Bakermans-Kranenburg, 2014), and other children
receiving nurturing care from foster caregivers that can play a critical role in promoting resilience for EB (Healey & Fisher, 2011;
Leve, Fisher, & Chamberlain, 2009). Although data on the quality of foster care were unavailable in our study, we encourage future
studies to incorporate these important factors when examining trajectories of EB for youth adopted from foster care.
Despite its lack of association with baseline EB in this sample, maltreatment did affect prospective change in EB during the first
five years post-adoption, and these associations differed by type of maltreatment. Sexual abuse predicted an increasing pattern of EB
in the first five years post-adoption, which is consistent with literature demonstrating the enduring effects of maltreatment on
biological and behavioral development. The specificity of these associations is consistent with other longitudinal samples identifying
sexual abuse as one of the strongest predictors of childhood EB (Nalavany et al., 2009 Nalavany, Ryan & Hinterlong, 2009; Simmel,
2007). These findings suggest that families adopting children from foster care with sexual abuse histories may need particularly

156
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

targeted support to help attenuate the pattern of EB growth in the first few years of adoption. In contrast, a history of family violence
exposure, such as witnessing domestic violence or sibling abuse, was associated with decreasing patterns of EB post-adoption. These
results suggest that children exposed to preadoption family violence may be particularly sensitive to positive changes in their
adoptive family environment, although follow-up studies are needed to investigate the specific mechanisms driving this pattern of
behavioral change.
Despite these associations in childhood, however, maltreatment did not predict long-term EB outcomes 11–15 years after
adoption, suggesting that preadoption maltreatment effects may be most relevant in the first few years of adoption. These findings
highlight the first few years of adoption as a critical period of behavioral change and plasticity. More studies are needed that focus on
identifying predictors of EB development (e.g., patterns of growth and desistance) across time. Just as increasing patterns of EB can
initiate cascades of negative developmental outcomes (Cicchetti & Banny, 2014; Rogosch et al., 2010), decreasing patterns of EB can
initiate trajectories of resilience consisting of improved academic, social, and global functioning (Masten, 2015). Future studies that
directly measure potential mediating factors (e.g., trauma symptoms, coping patterns) of EB desistance are also needed to elucidate
the causal mechanisms underlying the association between preadoption maltreatment and post-adoption EB.

4.2. Reactive temperament and EB

In contrast to theories of temperamental sensitivity to the environment, children with reactive temperaments did not show
heightened vulnerability to maltreatment. Similarly, they were marginally more sensitive to adoptive family cohesion with respect to
predictions of changes in EB across time, but this interaction did not reach statistical significance. Although the pattern of interaction
was consistent with the “bright side” of differential susceptibility, the direct effect of reactive temperament had a much more robust
association with EB in this high-risk sample of foster-adoptive youth. Specifically, reactive temperament was directly associated with
more severe EB at baseline, as well as with increasing patterns of EB in the first five years post-adoption, beyond the effects of
demographics, preadoption maltreatment, and adoptive family cohesion. The prospective association between reactive temperament
and later developmental patterns of EB is consistent with previous studies investigating temperamental risk underlying the etiology of
EB (Singh & Waldman, 2010). Given that reactive temperament (e.g., negative emotionality, low frustration tolerance) can be de-
tected as early as in early childhood and demonstrates sensitivity to changes in EB across time, identifying the early developmental
processes that contribute to the formation of reactive temperament may help to develop targeted prevention programs for children at
risk for EB.

4.3. Predicting adolescent/young-adult outcomes

Beyond examining temperament and environmental effects on EB in childhood, the final aim of our study was to explore how
temperament and positive changes in the family environment related to poor long-term outcomes for foster children: criminality and
substance use. Although reactive temperament robustly predicted EB and growth in EB across childhood, reactive temperament was
unrelated to EB outcomes in late-adolescence/young adulthood 11–15 years later. In fact, few of the early risk factors associated with
childhood EB extended to influence these long-term EB outcomes in adolescence. The transition from childhood to adolescence
includes many biological changes (e.g., hormonal fluctuations, brain development) as well as social changes, such as adolescents
spending more time away from the family context (Steinberg, 2008). In this study, adoptive family cohesion in childhood only
marginally predicted arrest history and was unrelated to substance use in adolescence/young adulthood, which is consistent with
evidence of the decreasing influence of the parental context in adolescence relative to peers and romantic partners (Laird, Jordan,
Dodge, Pettit, & Bates, 2001; Markiewicz, Lawford, Doyle, & Haggart, 2006). Future studies investigating pathways to adolescent EB
outcomes would benefit from adopting more proximal measures of adolescent environment, such as peer factors (e.g., peer deviance,
intimate relationships). Finally, it is important to note that despite all youth in our sample experiencing some significant early life
stress, their rates of arrest and substance use in late adolescence were comparable to normative populations (Brame, Turner,
Paternoster, & Bushway, 2012; Johnston, O’Malley, Miech, Bachman, & Schulenberg, 2017). These results support the overall pro-
tective role of adoption for children removed from their birth parents and placed into foster care.
The present findings should be interpreted in the context of several important study limitations. First, although our study is rare in
its longitudinal study of high-risk foster youth across seven time points spanning 11–15 years, the sample size (N = 82) may have
limited power to detect interactions; relatedly, although nonparticipants did not differ from participants in our study based on
demographic or other study variables, and missing data strategies (e.g., MICE) were employed to reduce bias, notable attrition and
missing data were observed across time. Future studies with larger samples must further assess temperament-based sensitivity to
maltreatment and family support across time, including evaluation of differential susceptibility. Second, shared method variance may
have influenced associations between parent-reported temperament, family-cohesion, and externalizing problems in childhood, al-
though we combined youth and parent-report whenever possible (i.e., at the young-adult follow-up). Finally, similar to other studies
of maltreatment (Appleyard, Berlin, Rosanbalm, & Dodge, 2011; Jonson-Reid et al., 2010), we used official records of abuse and
neglect to identify children with and without a history of maltreatment, which may reflect report bias and investigation bias. Fur-
thermore, although data about the chronicity and severity of maltreatment were unavailable in the current study, these factors are
critical to understanding the impact of maltreatment on behavioral development, and we encourage future studies of foster-adoptive
youth to incorporate these factors when exploring developmental pathways to EB.
Overall, these results highlight the importance of examining the longitudinal course of risk factors for youth adopted from foster
care, given that both early reactive temperament and pre-adoption maltreatment were associated with patterns of EB growth and

157
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

desistance in the first five years of adoption. We emphasize careful consideration of temperament when predicting EB development
for children in foster care, who are already at significant risk for EB across multiple stages of development. Compared to foster care
youth with easy temperaments, children with reactive temperaments had higher EB at baseline and were characterized by increasing
patterns of EB across the first five-years post-adoption, even with control of demographic variables (age of adoptive placement, sex,
race-ethnicity), maltreatment history, and adoptive family cohesion in childhood. Temperament was not purely a risk factor across
development, as youth with reactive temperaments were marginally more sensitive to family cohesion in the first five years of
adoption. Overall, these results highlight the importance of taking a developmental approach to prevention and intervention.
Findings suggest that a preventative approach to ameliorating EB may have the largest impact given the enduring predictions of
patterns of EB in childhood from reactive temperament and maltreatment. Reducing EB may require early interventions directly
targeted at temperamental factors underlying EB, such as negative emotionality or effortful control (DeLisi & Vaughn, 2014). Given
that few after-care services are currently offered and used by parents adopting children specifically from foster care, programs
tailored to foster children's unique needs (e.g., incorporating maltreatment subtype history and variations in temperament) may assist
with successful placements into permanent adoptive homes. For youth who already exhibit highly reactive temperaments and present
with elevated initial EB in their placements, it may be particularly important to take advantage of the heightened behavioral mal-
leability in early childhood by enhancing family support to maximize resilient pathways of development.

Acknowledgements

This research was partially supported by a National Science Foundation Graduate Research Fellowship awarded to Irene Tung; no
direct support was received from the National Science Foundation for this study's design, data analysis and interpretation, or writing
of the report. Special acknowledgement is due to Austin Blake and our research staff for their work in data collection and study
coordination, and we are indebted to the children and families who participated in our research study.

References

Achenbach, T. M., & Rescorla, L. (2001). ASEBA school-age forms & profiles. Burlington: Aseba.
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4-18 and 1991 profile. Burlington, VT: Department of Psychiatry, University of Vermont.
Appleyard, K., Berlin, L. J., Rosanbalm, K. D., & Dodge, K. A. (2011). Preventing early child maltreatment: Implications from a longitudinal study of maternal abuse
history, substance use problems, and offspring victimization. Prevention Science, 12(2), 139–149. http://dx.doi.org/10.1007/s11121-010-0193-2.
Barr, S. C., Hanson, R., Begle, A. M., Kilpatrick, D. G., Saunders, B., Resnick, H., & Amstadter, A. (2012). Examining the moderating role of family cohesion on the
relationship between witnessed community violence and delinquency in a national sample of adolescents. Journal of Interpersonal Violence, 27(2), 239–262. http://
dx.doi.org/10.1177/0886260511416477.
Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential susceptibility to environmental influences. Psychological Bulletin, 135(6), 885–908. http://dx.doi.
org/10.1037/a0017376.
Belsky, J., & Pluess, M. (2012). Differential susceptibility to long-term effects of quality of child care on externalizing behavior in adolescence. International Journal of
Behavioral Development, 36(1), 2–10. http://dx.doi.org/10.1177/0165025411406855.
Belsky, J., Hsieh, K.-H., & Crnic, K. (1998). Mothering, fathering, and infant negativity as antecedents of boys’ externalizing problems and inhibition at age 3 years:
Differential susceptibility to rearing experience. Development and Psychopathology, 10(02), 301–319.
Brame, R., Turner, M. G., Paternoster, R., & Bushway, S. D. (2012). Cumulative prevalence of arrest from ages 8 to 23 in a national sample. Pediatrics, 129(1), 21–27.
http://dx.doi.org/10.1542/peds.2010-3710.
Carey, W. B., & McDevitt, S. C. (1978). Revision of the infant temperament questionnaire. Pediatrics, 61(5), 735–739.
Cassidy, J., Woodhouse, S. S., Sherman, L. J., Stupica, B., & Lejuez, C. W. (2011). Enhancing infant attachment security: An examination of treatment efficacy and
differential susceptibility. Development and Psychopathology, 23(1), 131–148. http://dx.doi.org/10.1017/S0954579410000696.
Cicchetti, D., & Banny, A. (2014). A developmental psychopathology perspective on child maltreatment. In M. Lewis, & K. D. Rudolph (Eds.). Handbook of
Developmental Psychopathology (pp. 723–741). US: Springer. http://dx.doi.org/10.1007/978-1-4614-9608-3_37.
Courtney, M. E., Dworsky, A., Cusick, G. R., Havlicek, J., Perez, A., & Keller, T. (2007). Midwest evaluation of the adult functioning of former foster youth: Outcomes at age
21. Chicago, IL: Chapin Hall Center for Children at the University of Chicago. Retrieved from http://www.chapinhall.org/sites/default/files/
ChapinHallDocument_2.pdf.
Cutuli, J. J., Goerge, R. M., Coulton, C., Schretzman, M., Crampton, D., Charvat, B. J., et al. (2016). From foster care to juvenile justice: Exploring characteristics of
youth in three cities. Children and Youth Services Review, 67(Supplement C), 84–94. http://dx.doi.org/10.1016/j.childyouth.2016.06.001.
DeLisi, M., & Vaughn, M. G. (2014). Foundation for a temperament-based theory of antisocial behavior and criminal justice system involvement. Journal of Criminal
Justice, 42(1), 10–25. http://dx.doi.org/10.1016/j.jcrimjus.2013.11.001.
Edwards, A. C., Gardner, C. O., Hickman, M., & Kendler, K. S. (2016). A prospective longitudinal model predicting early adult alcohol problems: Evidence for a robust
externalizing pathway. Psychological Medicine, 46(5), 957–968. http://dx.doi.org/10.1017/S0033291715002457.
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2011). Differential susceptibility to the environment: An evolutionar-
y–neurodevelopmental theory. Development and Psychopathology, 23(01), 7–28. http://dx.doi.org/10.1017/S0954579410000611.
Euser, S., Alink, L. R. A., Tharner, A., van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2014). Out of home placement to promote safety? The prevalence of
physical abuse in residential and foster care. Children and Youth Services Review, 37(Suppl. C), 64–70. http://dx.doi.org/10.1016/j.childyouth.2013.12.002.
Fullard, W., McDevitt, S. C., & Carey, W. B. (1984). Assessing temperament in one-to three-year-old children. Journal of Pediatric Psychology, 9(2), 205–217.
Greeson, J. K. P., Briggs, E. C., Kisiel, C. L., Layne, C. M., Ake, G. S., Ko, S. J., et al. (2011). Complex trauma and mental health in children and adolescents placed in
foster care: Findings from the National Child Traumatic Stress Network. Child Welfare, 90(6), 91–108.
Hanley, J. A., Negassa, A., Edwardes, M. D., & Forrester, J. E. (2003). Statistical analysis of correlated data using generalized estimating equations: An orientation.
American Journal of Epidemiology, 157(4), 364–375.
Harris, K. M., Florey, F., Tabor, J., Bearman, P. S., Jones, J., & Udry, J. R. (2008). The national longitudinal study of adolescent health: Study design. Retrieved December 8,
2008, from http://www.cpc.unc.edu/projects/addhealth/design.
Haskett, M. E., Nears, K., Ward, C. S., & McPherson, A. V. (2006). Diversity in adjustment of maltreated children: Factors associated with resilient functioning. Clinical
Psychology Review, 26(6), 796–812. http://dx.doi.org/10.1016/j.cpr.2006.03.005.
Healey, C. V., & Fisher, P. A. (2011). Young children in foster care and the development of favorable outcomes. Children and Youth Services Review, 33(10), 1822–1830.
Hentges, R. F., Davies, P. T., & Cicchetti, D. (2015). Temperament and interparental conflict: The role of negative emotionality in predicting child behavioral problems.
Child Development, 86(5), 1333–1350. http://dx.doi.org/10.1111/cdev.12389.
Jaffee, S. R., Caspi, A., Moffitt, T. E., & Taylor, A. (2004). Physical maltreatment victim to antisocial child: Evidence of an environmentally mediated process. Journal of

158
I. Tung et al. Child Abuse & Neglect 76 (2018) 149–159

Abnormal Child Psychology, 113(1), 44–55.

Johnston, L. D., O’Malley, P. M., Miech, R. A., Bachman, J. G., & Schulenberg, J. E. (2017). Monitoring the Future national survey results on drug use, 1975–2016:
Overview, key findings on adolescent drug use. Ann Arbor: Institute for Social Research, The University of Michigan.
Jonson-Reid, M., Presnall, N., Drake, B., Fox, L., Bierut, L., Reich, W., et al. (2010). Effects of child maltreatment and inherited liability on antisocial development: An
official records study. Journal of the American Academy of Child & Adolescent Psychiatry, 49(4), 321–332. http://dx.doi.org/10.1016/j.jaac.2009.11.015.
Kerig, P. K., & Becker, S. P. (2015). Early abuse and neglect as risk factors for the development of criminal and antisocial behavior. In J. Morizot, & L. Kazemian (Eds.).
The development of criminal and antisocial behavior (pp. 181–199). Springer International Publishing. http://dx.doi.org/10.1007/978-3-319-08720-7_12.
Kiff, C. J., Lengua, L. J., & Zalewski, M. (2011). Nature and nurturing: Parenting in the context of child temperament. Clinical Child and Family Psychology Review, 14(3),
251–301.
Kliewer, W., Murrelle, L., Prom, E., Ramirez, M., Obando, P., Sandi, L., & Del Carmen Karenkeris, M. (2006). Violence exposure and drug use in Central American
youth: Family cohesion and parental monitoring as protective factors. Journal of Research on Adolescence, 16(3), 455–478. http://dx.doi.org/10.1111/j.1532-7795.
2006.00502.x.
Knudsen, E. I. (2004). Sensitive periods in the development of the brain and behavior. Journal of Cognitive Neuroscience, 16(8), 1412–1425. http://dx.doi.org/10.1162/
0898929042304796.
Kochanska, G., & Kim, S. (2013). Difficult temperament moderates links between maternal responsiveness and children's compliance and behavior problems in low-
income families. Journal of Child Psychology and Psychiatry, 54(3), 323–332.
Kuhlman, K. R., Geiss, E. G., Vargas, I., & Lopez-Duran, N. L. (2015). Differential associations between childhood trauma subtypes and adolescent HPA-axis func-
tioning. Psychoneuroendocrinology, 54(Suppl. C), 103–114.
Laird, R. D., Jordan, K. Y., Dodge, K. A., Pettit, G. S., & Bates, J. E. (2001). Peer rejection in childhood, involvement with antisocial peers in early adolescence, and the
development of externalizing behavior problems. Development and Psychopathology, 13(2), 337–354.
Leathers, S. J., Spielfogel, J. E., Gleeson, J. P., & Rolock, N. (2012). Behavior problems, foster home integration, and evidence-based behavioral interventions: What
predicts adoption of foster children. Children and Youth Services Review, 34(5), 891–899.
Leerkes, E. M., Nayena Blankson, A., & O’Brien, M. (2009). Differential effects of maternal sensitivity to infant distress and nondistress on social-emotional functioning.
Child Development, 80(3), 762–775. http://dx.doi.org/10.1111/j.1467-8624.2009.01296.x.
Leve, L. D., Fisher, P. A., & Chamberlain, P. (2009). Multidimensional treatment foster care as a preventive intervention to promote resiliency among youth in the child
welfare system. Journal of Personality, 77(6), 1869–1902.
Leve, L. D., Harold, G. T., Chamberlain, P., Landsverk, J. A., Fisher, P. A., & Vostanis, P. (2012). Practitioner review: Children in foster care –Vulnerabilities and
evidence-based interventions that promote resilience processes. Journal of Child Psychology and Psychiatry, 53(12), 1197–1211. http://dx.doi.org/10.1111/j.1469-
7610.2012.02594.x.
Li, C. (2013). Little's test of missing completely at random. Stata Journal, 13(4), 795–809.
Little, R. J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83(404),
1198–1202.
Markiewicz, D., Lawford, H., Doyle, A. B., & Haggart, N. (2006). Developmental differences in adolescents’ and young adults’ use of mothers, fathers, best friends, and
romantic partners to fulfill attachment needs. Journal of Youth and Adolescence, 35(1), 121–134.
Masi, G., Manfredi, A., Milone, A., Muratori, P., Polidori, L., Ruglioni, L., & Muratori, F. (2011). Predictors of nonresponse to psychosocial treatment in children and
adolescents with disruptive behavior disorders. Journal of Child and Adolescent Psychopharmacology, 21(1), 51–55. http://dx.doi.org/10.1089/cap.2010.0039.
Masten, A. S. (2015). Ordinary magic: Resilience in development. Guilford Publications.
Mcdevitt, S. C., & Carey, W. B. (1978). The measurement of temperament in 3–7 year old children. Journal of Child Psychology and Psychiatry, 19(3), 245–253.
Nalavany, B. A., Ryan, S. D., & Hinterlong, J. (2009). Externalizing behavior among adopted boys with preadoptive histories of child sexual abuse. Journal of Child
Sexual Abuse, 18(5), 553–573.
Odgers, C. L., Moffitt, T. E., Broadbent, J. M., Dickson, N., Hancox, R. J., Harrington, H., et al. (2008). Female and male antisocial trajectories: From childhood origins
to adult outcomes. Development and Psychopathology, 20(02), 673–716.
Oswald, S. H., Heil, K., & Goldbeck, L. (2010). History of maltreatment and mental health problems in foster children: A review of the literature. Journal of Pediatric
Psychology, 35(5), 462–472. http://dx.doi.org/10.1093/jpepsy/jsp114.
Palacios, J., & Brodzinsky, D. (2010). Review: Adoption research: Trends, topics, outcomes. International Journal of Behavioral Development, 34(3), 270–284.
Rioux, C., Castellanos-Ryan, N., Parent, S., Vitaro, F., Tremblay, R. E., & Séguin, J. R. (2016). Differential susceptibility to environmental influences: Interactions
between child temperament and parenting in adolescent alcohol use. Development and Psychopathology, 28(01), 265–275. http://dx.doi.org/10.1017/
S0954579415000437.
Rogosch, F. A., Oshri, A., & Cicchetti, D. (2010). From child maltreatment to adolescent cannabis abuse and dependence: A developmental cascade model. Development
and Psychopathology, 22(4), 883–897. http://dx.doi.org/10.1017/S0954579410000520.
Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147–177. http://dx.doi.org/10.1037/1082-989X.7.2.
147.
Simmel, C. (2007). Risk and protective factors contributing to the longitudinal psychosocial well-being of adopted foster children. Journal of Emotional and Behavioral
Disorders, 15(4), 237–249. http://dx.doi.org/10.1177/10634266070150040501.
Singh, A. L., & Waldman, I. D. (2010). The etiology of associations between negative emotionality and childhood externalizing disorders. Journal of Abnormal
Psychology, 119(2), 376–388. http://dx.doi.org/10.1037/a0019342.
Steinberg, L. (2008). A social neuroscience perspective on adolescent risk-taking. Developmental Review, 28(1), 78–106. http://dx.doi.org/10.1016/j.dr.2007.08.002.
Sterne, J. A. C., White, I. R., Carlin, J. B., Spratt, M., Royston, P., Kenward, M. G., et al. (2009). Multiple imputation for missing data in epidemiological and clinical
research: Potential and pitfalls. BMJ (Clinical Research Ed.), 338, b2393.
U.S. Department of Health and Human Services, Trends in foster care and adoption, FY2002-FY 2010, 2011, Retrieved from http://www.acf.hhs. gov/programs/cb/
stats_research/afcars/trends_june2011.pdf.
van IJzendoorn, M. H., & Bakermans-Kranenburg, M. J. (2012). Differential susceptibility experiments: Going beyond correlational evidence: Comment on beyond
mental health, differential susceptibility articles. Developmental Psychology, 48(3), 769–774.
van IJzendoorn, M. H., & Juffer, F. (2006). The Emanuel Miller Memorial Lecture 2006: Adoption as intervention. Meta-analytic evidence for massive catch-up and
plasticity in physical, socio-emotional, and cognitive development. Journal of Child Psychology and Psychiatry, 47(12), 1228–1245. http://dx.doi.org/10.1111/j.
1469-7610.2006.01675.x.
Vidal, S., Prince, D., Connell, C. M., Caron, C. M., Kaufman, J. S., & Tebes, J. K. (2017). Maltreatment, family environment, and social risk factors: Determinants of the
child welfare to juvenile justice transition among maltreated children and adolescents. Child Abuse & Neglect, 63(Suppl. C), 7–18. http://dx.doi.org/10.1016/j.
chiabu.2016.11.013.
Welsh, B. C., Loeber, R., Stevens, B. R., Stouthamer-Loeber, M., Cohen, M. A., & Farrington, D. P. (2008). Costs of juvenile crime in urban areas: A longitudinal
perspective. Youth Violence and Juvenile Justice, 6(1), 3–27. http://dx.doi.org/10.1177/1541204007308427.

159